INFORMATION
ECONOMICS
AND POLICY
Information Economics and Policy 19 (2007) 379–404
www.elsevier.com/locate/iep
Bundling, product choice, and efficiency:
Should cable television networks
be offered à la carte?
Gregory S. Crawford *, Joseph Cullen
Department of Economics, University of Arizona, United States
Received 27 September 2006; received in revised form 12 June 2007; accepted 18 June 2007
Available online 6 July 2007
Abstract
We conduct a numerical analysis of bundling’s impact on a monopolist’s pricing and product
choices and assess the implications for consumer welfare in cable television markets. Existing theory
is ambiguous: for a given set of products, bundling likely transfers surplus from consumers to firms
but also encourages products to be offered that might not be under à la carte pricing. Simulation of
‘‘Full À La Carte’’ for an economic environment calibrated to an average cable television system
suggests that consumers would likely benefit from à la carte sales. If all networks continued to be
offered, the average household’s surplus is predicted to increase by $6.80 (65.6%) under à la carte
sales (despite a total bundle price that almost doubles) and reduced network profits would have
to be such that 41 of 50 offered cable networks have to exit the market to make her indifferent. Simulation of a ‘‘Theme Tier’’ scenario provides intermediate benefits. The incremental marginal costs to
cable systems of à la carte sales and its impact in the advertising market and on competition are
important factors in determining consumer benefits.
2007 Elsevier B.V. All rights reserved.
JEL classification: L12; L82; L50
Keywords: Bundling; Product choice; Cable television; à la carte; Policy analysis
*
Corresponding author. Tel.: +1 520 621 5247.
E-mail address: [email protected] (G.S. Crawford).
0167-6245/$ - see front matter 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.infoecopol.2007.06.005
380
G.S. Crawford, J. Cullen / Information Economics and Policy 19 (2007) 379–404
1. Introduction
Bundling is a common feature in many imperfectly competitive product markets. Firms
in the telecommunications, financial services, health care, and information industries frequently offer products in bundles. While often innocuous, recent research has identified
settings in which bundling can be used by firms to price discriminate among consumers
or to extend market power into related product markets (Adams and Yellen, 1976; Bakos
and Brynjolfsson, 1999; Nalebuff, 2004).
We conduct a numerical analysis of bundling’s impact on the prices and networks
offered by a ‘‘typical’’ US cable television system and assess the consequences of these
choices for consumer, producer, and total welfare. We do so to speak to the current public
policy debate about bundling in cable markets. Motivated by consumer dis-satisfaction
over ever-increasing cable prices, law- and policy-makers are looking at unbundling –
or ‘‘à la carte’’ pricing – as a possible policy solution.1 Most cable television systems
and the networks provided on them are strongly opposed to the idea, in part claiming that
doing so would (possibly dramatically) reduce the number of (especially smaller) networks
that can survive in an à la carte environment (Booz Allen Hamilton, 2004).2
Existing theory bears out both of these views. On bundling and pricing, a substantial
theoretical literature suggests that bundling may be used to sort consumers in a manner
similar to second-degree price discrimination (Stigler, 1968; Adams and Yellen, 1976;
McAfee et al., 1989; Bakos and Brynjolfsson, 1999; Armstrong, 1999). When consumers
have heterogeneous tastes for several products, bundling reduces that heterogeneity,
allowing the firm to earn greater profit than would be possible with component (unbundled) prices. While firms clearly benefit in this case, consumer welfare generally falls, particularly when bundling requires consumers to purchase products in which they have little
interest.
Less well established are the implications of bundling on the products offered by firms.
It is well known that a single-product monopolist will not offer some products that
increase social welfare because of the non-appropriability of total surplus (e.g. Tirole,
1988, Chapter 2). A key factor in the extent of the distortion is the shape of consumer preferences; all else equal, products with more elastic demand are more likely to be offered as
the monopolist can better appropriate surplus for such goods. In the multi-product case,
theory is ambiguous: under-provision due to non-appropriability is still a problem, but if
products are substitutes, monopoly pricing on one good may increase demand for a second enough to encourage its (inefficient) offering. In this (multi-product) case, bundling
can also influence the calculus of product choice. Because bundling aggregates preferences
for bundle components, it makes tastes more homogenous, increasing the elasticity of
demand (Bakos and Brynjolfsson, 1999; Crawford, 2006). If marginal costs for bundle
components are zero, Bakos and Brynjolfsson (1999) show that as bundle size increases
without bound, the product choice problem is solved; bundle profit converges to total surplus and both the monopolist and social planner will offer the same portfolio of goods.
1
In addition to numerous articles in the popular press (e.g. Reuters, 2003; Squeo and Flint, 2004; Shatz, 2006),
the Federal Communications Commission (FCC) has published two reports analyzing à la carte pricing (FCC,
2004, 2006).
2
The National Cable and Telecommunications Association (NCTA) has a useful webpage summarizing
industry perspectives at http://www.ncta.com/IssueBrief.aspx?contentId=15.
G.S. Crawford, J. Cullen / Information Economics and Policy 19 (2007) 379–404
381
When marginal costs are not zero, however, bundling can work too well. Because bundling requires all consumers purchase all goods, some sales may arise to consumers that
value components at less then their cost. This is particularly problematic when there is
strong negative correlation in tastes. Bundling is especially profitable in such settings
(Adams and Yellen, 1976; Schmalensee, 1984) and the profit gains from correlation may
outweigh the profit losses from below-cost sales. If so, the bundling monopolist may offer
products that actually reduce total welfare. Even if not, consumer welfare may be lower
from bundling; while some consumers will benefit from a new product, these gains may
be outweighed by losses to existing consumers of the bundle.3
What are the implications of these results for the à la carte debate in cable markets? The
first branch of the literature strongly suggests that, for a given set of networks, bundling
reduces consumer welfare to the benefit of firms. The second branch, however, suggests
that by providing stronger incentives to offer new products, bundling may increase consumer (and total) welfare. The natural solution is to attempt to address this question
empirically, a topic of related work in progress (Crawford and Yurukoglu, 2007). In this
paper, we take a numerical approach. We make assumptions about households’ willingness-to-pay (WTP) and firms’ costs for television networks. We then calibrate these
assumptions to replicate the offerings of an ‘‘average’’ 2004 cable system (California Cable
of Monterey, CA) and simulate the profit and welfare consequences of alternative à la
carte policy proposals. We do not consider the impact of à la carte on competition and
the advertising market (two important caveats), but we do consider the consequence of
increased marginal costs to cable systems of à la carte offerings.
Several interesting results emerge from our numerical simulations. We consider two à la
carte scenarios: ‘‘Full À La Carte’’ and ‘‘Theme Tiers’’. As expected, if all networks continue to be offered, consumers gain and firms lose from Full À La Carte: average perhousehold consumers surplus is estimated to increase by $6.80 (a 65.6% increase), gross
firm profit is estimated to fall by $9.08 (a 44.2% decrease), and gross total surplus is estimated to fall by $2.28 (a 7.4% decrease). These impacts almost surely outweigh any compensating benefit of à la carte to enhance incentives to offer networks: we estimate that 41
of the 50 offered cable networks in Monterey would have to exit for consumers surplus
under Full À La Carte to be no higher than that under bundling. Theme tiers offer similar,
though smaller, effects (a 14.2% increase in consumers surplus, a 15.7% decrease in profit,
a 5.7% decrease in total surplus, and 24 networks needed to exit to equate consumer benefits). These results suggest consumers might well benefit from à la carte sales of cable
networks.
3
In addition to the general literature cited here, there is a sizable literature analyzing the programs offered by
firms in media (esp. radio and television) markets. Dating to Steiner (1954), this literature has focused on the
consequences of competition (versus monopoly) and advertiser (versus pay) support on the number and type of
programs offered by firms and their consequence to consumer, advertiser, and total welfare (cf. Owen and
Wildman, 1992; Doyle, 1998; Berry and Waldfogel, 1999; Anderson and Coate, 2005). The contribution of this
paper is to focus on the case most relevant to cable television – pay-supported multi-product monopoly – and
analyze the impact of bundling on firms’ choices. Extending the analysis here to consider the impact of
competition with satellite systems and the advertising market would be valuable contributions. We discuss the
likely consequences of these omissions in Section 3.
382
G.S. Crawford, J. Cullen / Information Economics and Policy 19 (2007) 379–404
2. The welfare effects of bundling
In this section we illustrate the welfare effects of bundling by a multi-product monopolist. In the first subsection, we illustrate the discriminatory effects of bundling for a given
set of products. In the second, we extend this result to consider the impact of bundling on
the set of offered products.
2.1. The discriminatory effects of bundling
2.1.1. The case of two goods
Most of the discriminatory bundling literature has focused on the incentives to bundle
two goods. Adams and Yellen (1976) formalize the seminal work of Stigler (1963) and
present examples where bundling is more or less profitable than component (unbundled)
sales. Schmalensee (1984) and Salinger (1995) extend the analysis to the case of normal
and uniform tastes. A simple example, adapted from Adams and Yellen (1976), demonstrates the discriminatory incentives.
There are two goods (1, 2) and four consumers (A–D), whose willingness-to-pay (WTP)
for each good is represented by a point in the top panel of Fig. 1. The bottom three panels
show the demand for each good (if offered separately) and demand for the bundle of both
goods implied by these reservation values. Marginal costs are c1 = 20 for good 1 and
c2 = 30 for good 2.
Unbundled sales imply optimal prices of p1 ¼ 60 and p2 ¼ 90, yielding consumers surplus (in market 1) of $30 and (combined) profits of $140. Consumers surplus and profits in
each market (if any) are labelled by CS and P in the figure. Under bundling, however, perfect negative correlation in tastes for each component imply that all consumers have WTP
of $100, yielding bundled profits of $200. In this example, bundling permits the monopolist to extract all available consumers surplus.
The reduction in preference heterogeneity in the example (and associated surplus extraction) generalizes and is the primary benefit of bundling. It is not sufficient, however. In a more
general setting, when bundled sales are preferred to component sales depends on three critical features of preferences and costs. First is the extent of heterogeneity reduction possible
from bundling. This increases with the negative correlation in preferences for bundle components, a point made clear by the example.4 Second is the level of marginal costs for components. Since bundling requires that consumers purchase all goods, some below-cost sales of
components can result (e.g. consumers A and D in the example), reducing the gains from
bundling. Below-cost sales become more likely the higher marginal costs are relative to the
mass of consumer preferences.5 Third is that bundling requires firms charge a single price.
When consumer tastes for components differ considerably (e.g. multiply WTP for one of
the example goods by 100), bundling is less attractive than component sales as it permits
fewer instruments (prices) to capture consumers’ surplus.6
4
Negative correlation, however, is not necessary for bundling to be profitable (McAfee et al., 1989).
Evans and Salinger (2005) further discuss cost-based arguments for bundling.
6
McAfee et al. (1989) extend the analysis of Adams and Yellen (1976) to consider mixed bundling, the offering
of both component and bundled sales, and show it always yields (weakly) greater profits than pure bundling. The
reason for this is clear: it maintains the benefits of bundling (if any) and strictly increases the number of prices
available to capture surplus. Despite this fact, mixed bundling is relatively uncommon, perhaps due to the added
administrative costs associated with offering both bundled and component goods.
5
G.S. Crawford, J. Cullen / Information Economics and Policy 19 (2007) 379–404
383
Reservation Values
100
A
R
2
90
B
55
C
40
c = 30
2
D
10
0
0
10
45
c1 = 20
Demand for Good 1
120
60
R1
Demand for Good 2
90
Demand for Bundle
120
120
*
b
100
*
p2 =
80 CS
100
p = 100
90
80
Πb
1
*
p1 =
60 Π
2
60
Π1
40
cb = 50
c 2 = 30
c = 20
20
1
0
0
1 q* = 2 3
1
4
5
0
0 q* = 1 2
2
3
4
5
0
0
1
2
3 q* = 4 5
b
Fig. 1. An example of the discriminatory incentives to bundle.
2.1.2. More than two goods
Recent papers by Bakos and Brynjolfsson (1999) and Armstrong (1999) extend the analysis of bundling to consider multiple goods. If there is no substitutability or complementarity in demand, consumer WTP for bundles is the simple sum of the WTP for the
components. Assuming the monopolist cannot observe any individual consumer’s WTP
but knows the distribution of WTP in the population, he can construct the optimal product
384
G.S. Crawford, J. Cullen / Information Economics and Policy 19 (2007) 379–404
2 Goods
20 Goods
20
0.9
1.8
18
0.8
1.6
16
0.7
1.4
14
0.6
1.2
12
0.5
Price
2
Price
Price
1 Good
1
1
10
0.4
0.8
8
0.3
0.6
6
0.2
0.4
4
0.1
0.2
2
0
0
0
0.5
Market Share
1
0
0
0.5
Market Share
1
0
0.5
Market Share
1
Fig. 2. Bundling homogenizes preferences, flattening demand.
line. While it may appear to be sub-optimal for him to offer a bundle at a fixed fee, the
optimal tariff can be quite complex and difficult to calculate, even for simple preference
structures (Armstrong, 1996; Rochet and Chone, 1998; Rochet and Stole, 2000).7
The consequences of bundling for the distribution of consumer WTP is significant. Bundling aggregates (averages) consumer tastes for the bundle components, xj. When bundles
are large, a Law of Large Numbers (LLN) effect obtains: the distribution of preferences
for the bundle becomes more concentrated as n increases (e.g., White, 1984). The implication of this result for the bundle demand curve is demonstrated in Fig. 2, adapted from
Bakos and Brynjolfsson (1999). For the case of uniformly distributed tastes (i.e. linear
demand for components), the figure presents the demand curve for a bundle of size 1,
2, and 20.8 As bundle size increases, there are fewer extreme tastes, corresponding to an
increasingly flat demand curve and greater consumer surplus extraction.9 Fig. 1 exhibited
a similar effect for just two goods.
Does the monopolist benefit from this reduction in heterogeneity? As in the two-good
case, it does when costs are low (discouraging below-cost sales of components) and when
tastes are not too extreme (which favors pricing components separately).
2.1.3. The welfare effects of discriminatory bundling
What are the implications of discriminatory bundling for welfare? When costs are zero,
Bakos and Brynjolfsson (1999, Proposition 1) find that as bundle size increases, both con7
Armstrong (1999) shows that the proportion of first-best profits obtainable by bundling is given by
p3 ffiffi Þ. This implies for a bundle of a size common in the cable television industry (e.g. between
P ðn þ 1Þ1=n ð1 0:97
n
30 and 60), a simple fixed fee tariff yields profits of at least 77–81% of the first-best profit.
8
Similar effects obtain for other distributions.
9
This may seem counter-intuitive. For a fixed level of demand (e.g. rotate a linear demand curve around its
intersection with the quantity axis), a monopolists profit is higher the more inelastic is demand. Bundling,
however, simultaneously shifts out and flattens the aggregate demand curve.
p
p
G.S. Crawford, J. Cullen / Information Economics and Policy 19 (2007) 379–404
385
sumers surplus and deadweight loss per good converge to zero.10 As bundle size grows
without bound, firms extract all the surplus available in the market. On average, consumers lose, but firms gain by more than consumers lose and total surplus increases.11
When marginal costs are positive, bundling becomes less desirable for all of consumers,
firms, and society. Bundling with positive marginal costs admits the possibility of belowcost sales, reducing gains to all parties. Indeed, as we show in the next section, it is possible
for a monopolist to have the incentive to profitably introduce products that actually
reduce total welfare. Regardless, positive costs does not attenuate the central lesson of this
literature: bundling can help a monopolist extract consumers surplus and increase profit.
2.2. Bundling and product choice
The previous section took the set of offered products as given. This is inappropriate if
bundling itself can increase the set of offered products, to the benefit of consumers, firms,
and society as a whole. To address this issue, we therefore consider the incentives facing a
monopolist to introduce new goods on either a stand-alone or bundled basis.
2.2.1. Primitives
Suppose a monopolist currently offers a single product and is considering the offer of an
additional product either as a stand-alone (à la carte) product or in a bundle with his existing product.
Let p1 and p2 be the prices charged by the monopolist for each product if he offers the
second on an à la carte basis and let p1+2 be the price charged if he offers them as a bundle.
Let w1(p1) = D1(p1)/N, w2(p2) = D2(p2)/N, and w1+2(p1+2) = D1+2(p1+2)/N be the associated per-capita demand curves and market shares for known market size N. Let the associated per-capita variable cost functions be Ci(wi) = ciwi, for i = {1, 2, 1 + 2}, where ci is
the marginal cost for product i. At times, we will also allow for per-capita fixed costs,
Fi, for offering each product.
Let per-capita consumer surplus, gross profit (i.e. ignoring fixed costs), and gross total
surplus for each product, i = {1, 2, 1 + 2}, be given by
Z 1
S i ðpi Þ ¼
wi ðvÞdv
ð1Þ
p
pi ðpi Þ ¼ pi wi ðpi Þ C i ðwi ðpi ÞÞ ¼ ðpi ci Þwi ðpi Þ
ð2Þ
W i ðpi Þ ¼ pi ðpi Þ þ S i ðpi Þ
ð3Þ
Let pM
i ¼ arg maxpi pi ðp i Þ be the Profit-maximizing price charged by the monopolist for
each product and define the associated (gross) profit, consumer, and total surplus as
M
M
M
M
M
pM
i ¼ pi ðp i Þ; S i ¼ S i ðp i Þ, and W i ¼ W i ðpi Þ. Similarly let p i ¼ arg maxpi W i ðp i Þ be the
welfare-maximizing price for each product with associated (gross) profit, consumer, and
total surplus, pi , S i , and W i defined analogously.
10
The assumptions underlying this result are three: (1) zero marginal costs, (2) independent preferences for
components, and (3) free disposal. The second of these can be relaxed as long as preferences for components are
not too positively correlated. See Bakos and Brynjolfsson (1999) and Crawford (2006).
11
Throughout this paper, we take total surplus to be the simple sum of consumers and producers surplus. Other
social welfare functions (e.g. a Rawlsian one that maximizes the minimum of the two) would yield very different
conclusions.
386
G.S. Crawford, J. Cullen / Information Economics and Policy 19 (2007) 379–404
Exponential Demand Curve, μ = 1
3.5
M
*
β = π2 /W2 = 0.37
3
Price
2.5
2
A
1.5
pM
=
$1.31
2
1
B
C
0.5
c = $0.30
2
0
0
0.1
0.2
0.3
0.4
0.5
0.6
Market Share
0.7
0.8
0.9
1
Fig. 3. The incentives to offer stand-alone products.
2.2.2. Incentives for stand-alone products
Assume that there is no substitutability or complementarity in demand between the two
products.12 If the monopolist offers the potential new product on an à la carte basis, his
incremental profit is just the profit from that product, pM
2 .
The incentives facing the monopolist in this case can be summarized by Fig. 3.
Displayed is the demand curve corresponding to an exponential distribution of willingness-to-pay (WTP) for the potential product in the population of consumers. The mean
and variance of this distribution are both equal to one. Also reported are the profit-maximizing price and market share for this demand curve under the assumption that marginal
costs are $0.30.
Profit-maximizing gross profit, pM
2 , is given by the area B in the figure. Welfare-maximizing gross total surplus, W 2 , is given by the sum of the areas A–C. If there are fixed costs
from introducing the new product, any fixed cost between B and (A + B + C) will prevent
the monopolist from introducing the product on an à la carte basis even though it could
increase social welfare. The reason is simple: the monopolist cannot appropriate the total
12
All else equal, substitutability (complementarity) will increase (decrease) the monopolist’s incentives to offer
the product.
G.S. Crawford, J. Cullen / Information Economics and Policy 19 (2007) 379–404
Exponential Demand Curve, μ = 1
Bundle Demand Curve, Each Exponential μ = 1
πM = 0.27
5
ΔπM = 0.30
5
1
M
1
M
*
1
β = π /W = 0.37
*
Δβ = Δπ /W2 = 0.41
4
Price
4
Price
387
3
3
A1+2
M
= $2.08
p1+2
2
A
1
M
1
p = $1.29
2
B
1
c = $0.30
B1+2
C
1+2
c1+2= $0.60
C1
1
0
0
0.2 0.4 0.6 0.8
Market Share
1
0
0
0.2 0.4 0.6 0.8
Market Share
1
Fig. 4. The incentives to offer products in a bundle.
surplus from the introduction of the product. It loses some to consumers surplus (A) and
some to the deadweight loss from monopoly pricing (C).
Spence (1975) introduced b ” pM/W* to measure the difference between private and
social incentives to offer a new product.13 b 2 [0, 1] and the larger it is, the more the
monopolist can appropriate the total potential surplus from introducing the product.
For exponential tastes, the monopolist can only capture 37% of the total surplus created
by offering the product.
2.2.3. Incentives for bundles
Now consider the case where the monopolist offers the new product as part of a bundle
with his existing product. For simplicity, we assume consumer preferences and marginal
costs for each product are the same as in Fig. 3. Fig. 4 describes the consequences of bundling on the monopolist’s incentives.
Pictured in the left panel is the demand for the monopolist’s existing product when
WTP is distributed as an exponential in the population of households. Pictured in the right
panel is the demand for a bundle of the two products when WTP for each is distributed as
an exponential.
Bundling changes the calculus of product choice. When bundling, the monopolist comM
pares the incremental profit of adding the good to his existing bundle ðDpM ¼ pM
1þ2 p1 ¼
B1þ2 B1 Þ to any fixed costs he might incur. From an efficiency perspective, the question is
13
Technically he introduced the idea to measure the difference between private and social incentives for selecting
a quality for a product necessarily offered by the monopolist, but the incentives in both cases are similar.
388
G.S. Crawford, J. Cullen / Information Economics and Policy 19 (2007) 379–404
whether this change in profit is more or less than the incremental surplus from offering the
product on an à la carte basis, i.e. whether DpM is closer or further to W 2 .14 Let Db ¼
DpM =W 2 . The figure demonstrates that bundling can help solve the product choice problem. Comparing b and Db, the figure shows that for the exponential distribution, adding
the second product to the bundle increases the share of the surplus the monopolist can
capture from that product.
Bundling also changes the welfare consequences of additional products. In the à la carte
case, whenever a firm introduces a product it necessarily increases both consumer and
total welfare (cf. Fig. 3). There is no such guarantee when adding a product to a bundle.
M
Adding a new product can decrease consumer welfare ðDS M S M
1þ2 S 1 Þ, decrease total
M
M
M
welfare ðDW W 1þ2 W 1 Þ or both. The reason is that bundling enhances profitability
by exploiting consumer preferences across goods. It may therefore be profitable to add a
new product to a bundle solely to increase surplus extraction on other products, even if
adding the new product decreases total welfare. Similarly, even if some consumers benefit
from the addition of a new product to the bundle, their gains may be outweighed by losses
to existing consumers of the bundle.
2.2.4. Mean preferences, marginal costs, and correlation
What factors influence when bundling enhances incentives to offer new products? When
it reduces welfare? As earlier, two key factors determine the consequences of bundling on
profits and welfare: the difference between marginal cost and mean WTP for components
and correlation in that WTP for components.15
To explore the effect of each of these factors, consider again a monopolist that currently
offers a bundle and is considering offering a new product, either on an à la carte basis or as
part of his existing bundle. Suppose for convenience he could select among a continuum of
potential products that differ in several dimensions: the distribution, mean, and variance of
WTP for the product, the marginal and fixed costs the monopolist would have to pay, and
the degree of correlation in household tastes between the new product and the existing
bundle.16
Fig. 5 demonstrates the influence of the difference between mean WTP and marginal
costs on the profit and welfare of bundling while Fig. 6 demonstrates the influence of negative correlation on the profit and welfare of bundling. In each figure, 4 panels are shown.
The top two panels in each figure present the efficiency consequences of bundling (i.e.
‘‘How close does bundling get to ensuring offerings that could increase total welfare?’’).
Reported are the incremental profit from bundling (DPM) or à la carte sales ðPM
2 Þ as well
as the total surplus available from offering it on an efficient à la carte ðW 2 Þ or bundled
(DW*) basis.17 The bottom two panels in each figure present the actual welfare conse14
An alternative efficiency criterion would be to compare the change in profit with the change in welfare from
offering the product in the bundle (i.e. Dp* versus DW*). While reasonable (and the approach taken by Anderson
and Coate, 2005), this appears further from the original intent of the product choice problem.
15
Schmalensee (1984) establishes the importance of these factors in a numerical analysis with two goods. Fang
and Norman (2005) confirm the importance of the first in general settings.
16
In particular, let preferences for the bundle be distributed normally with mean WTP of 30, a standard
deviation of 8, and let marginal costs for the bundle be 9. Unless otherwise stated, let mean WTP for the potential
new product be a truncated normal with mean 1 and standard deviation 1 and let its marginal cost be 0.30. These
values are not critical; they were chosen to facilitate identifying the influence of each of the factors.
17
That is, the total surplus when setting price equal to marginal cost.
G.S. Crawford, J. Cullen / Information Economics and Policy 19 (2007) 379–404
c =0
389
c =1
2
2
2.5
1.5
W*
2
2
1
πM
Δ W*
1.5
Dollars
Dollars
2
Δπ M
1
0.5
0
0.5
0
-0.5
0
0.5
1
μ
1.5
-1
2
0
0.5
2
c =0
1
μ2
1.5
2
1.5
2
c =1
2
2
2
1
WM
2
S2M
0.5
Δ WM
ΔS
1
Dollars
Dollars
1.5
M
0.5
0
0
-0.5
0
0.5
1
μ2
1.5
2
-1
0
0.5
1
μ2
Fig. 5. The Importance of marginal costs relative to mean preferences on À-La-Carte versus bundling outcomes.
quences of bundling (i.e. ‘‘What actually happens to consumer and total surplus from bundling?’’). Reported are the incremental total and consumers’ surplus from à la carte and
M
M
M
bundled sales (W M
2 , S 2 , DW , and DS ). Of course, actual welfare (in the bottom panels)
is always less than potential welfare (in the top panels) due to the distortions caused by
monopoly pricing.
Fig. 5 demonstrates the importance of marginal costs for the profit and welfare of à la
carte versus bundled sales. When costs are zero (left-hand panels), total potential welfare
from bundled and component sales are equal (W 2 DW ) and bundling yields greater
profit. For any potential product, therefore, there are a range of fixed costs for which that
product would be offered in a bundle and not offered on an à la carte basis and total welfare would increase. Consumers, too, would benefit from the offering. When costs are positive, however, whether bundling could be more efficient than à la carte sales depends on
the level of mean WTP for the product. For low levels of mean WTP (i.e. marginal costs
higher than mean tastes), bundling is both profit- and welfare-reducing. For higher levels
of mean WTP, however, bundling does better, providing greater profit (and thus stronger
390
G.S. Crawford, J. Cullen / Information Economics and Policy 19 (2007) 379–404
μ = 0.75, c2 = 0.5
0.6
0.6
0.4
0.4
Dollars
Dollars
μ = 0.75, c2 = 1.0
0.2
-0.2
-0.2
0
ρ
-0.4
0.5
0.6
0.6
0.4
0.4
0.2
-0.2
-0.2
ρ
0
ρ
0.5
0.5
M
2
SM
2
W
Δ WM
0
0
-0.5
0.2
0
-0.5
ΔπM
μ = 0.75,c2 = 0.5
Dollars
Dollars
μ = 0.75,c2 = 1.0
-0.4
2
0.2
0
-0.5
πM
Δ W*
0
-0.4
*
W2
-0.4
Δ SM
-0.5
0
0.5
ρ
Fig. 6. The importance of correlation and cost on À-La-Carte versus bundling outcomes.
incentives for product introductions) when tastes are approximately 40% greater than
costs. Note also in the bottom panels that consumers’ surplus from bundling is everywhere
lower than that from à la carte sales.
Fig. 6 demonstrates the importance of negative correlation (and costs) for the profitability
of bundled sales. In all the panels, the correlation between the existing bundle and the new
product is allowed to vary between 0.7 and 0.7.18 In the left-hand (right-hand) panels, mean
M
M
tastes are slightly below (above) costs. While à la carte profit and welfare ðPM
2 ; S 2 andW 2 Þ
M
are invariant to correlation, the profit and welfare consequences of bundling (DP , DSM,
and DWM) are highly dependent on it. For large negative correlation and costs high relative
to tastes (the left-hand panels), consumer and total surplus can actually fall with the bundling
of new goods. This cannot happen under à la carte sales. The right-hand panel demonstrates
the importance of high marginal costs to this conclusion. It duplicates the left-hand panel
18
This is a high but not unreasonable level of correlation. It is possible to generate correlations of this order if
one could identify a potential product that appealed (exclusively) to the bottom 10–11% of the population of
tastes for the existing bundle.
G.S. Crawford, J. Cullen / Information Economics and Policy 19 (2007) 379–404
391
after cutting marginal costs in half. While profitability is still strongly decreasing in correlation, bundling never reduces welfare. It is therefore the interaction of high costs and negative
correlation that can yield bundling that reduces total welfare.
3. Bundling, product choice, and welfare in cable television markets
The previous section demonstrated that bundling generally transfers surplus from consumers to firms, but can also encourage the introduction of products that would not otherwise be offered. This section applies these ideas to the cable television industry. Cable
television systems choose a portfolio of television networks, bundle them into services,
and offer these services to consumers in local, geographically separate, cable markets.
The three main types of program networks offered on cable are broadcast networks, cable
networks, and premium networks.19 Broadcast and cable networks are typically bundled
and offered as Basic and Expanded Basic Services while premium programming networks
are typically unbundled and sold as Premium Services.
The bundling of cable networks into Basic and Expanded Basic services have recently
come under scrutiny by both law- and policy-makers. Frustrated by ever-increasing prices,
consumers and their representatives in the federal government have asked if unbundled
sales might not be a potential solution. If consumers select just those channels that they
prefer, might not cable bills fall? Cable systems and the television networks offered on
them have been vocal in their opposition. They argue that à la carte sales would (i)
increase costs, (ii) decrease advertising revenues, and (iii) reduce the viability of (especially
small, niche) networks.
To address these questions, we calibrate the model above to an ‘‘average’’ system in the
cable television industry and simulate the effects of two à la carte proposals considered in
the popular press. In so doing, we analyze the pricing and bundling for this system’s most
popular Expanded Basic service. A recent survey of cable television systems conducted by
the Federal Communications Commission (FCC) found that the average US cable system
charged $45.32, had a market share of 72%, and offered 45 Expanded Basic channels
(FCC, 2005a,b). Surveying the population of cable systems from Warren (2005), we
selected California Cable of Monterey, California as our sample system.20 This system’s
most popular Expanded Basic service offered 50 cable networks, charged a price of
$44.05, and had a market share of 74%.
We call our two à la carte scenarios ‘‘Full À La Carte’’ and ‘‘Theme Tiers’’. In both proposals we assume that before households can purchase any networks on an à la carte basis,
they must first purchase a ‘‘Limited Basic’’ bundle consisting of all available broadcast and
public interest channels offered in Monterey. For analytical convenience, we assume that all
households value this bundle at $5 and that California Cable of Monterey therefore charges
a price of $5 for it.21 In the ‘‘Full À La Carte’’ scenario, we assume that California Cable of
19
Examples of each type are ABC, CBS, NBC, and FOX for broadcast networks, MTV, CNN, and ESPN for
cable programming networks, and HBO and Showtime for premium networks.
20
California Cable is owned by Comcast, the largest Multiple System Operator (MSO) in the United States.
21
Modelling household preferences for broadcast networks is more challenging that for cable networks as cable
systems do not generally receive payment for them. Instead, the owners of television networks generally negotiate
carriage agreements for their affiliated cable networks. Furthermore, modelling the demand for and pricing of à la
carte networks in the presence of heterogeneous tastes for the ‘‘Limited Basic’’ bundle is much more complicated.
See Crawford and Yurukoglu (2007) for a detailed discussion of this issue.
392
G.S. Crawford, J. Cullen / Information Economics and Policy 19 (2007) 379–404
Monterey offers every current cable network in the most popular Expanded Basic bundle on
an à la carte basis instead. In the ‘‘Theme Tier’’ scenario, we assume that they instead bundle
networks with similar subject matter into 11 tiers and offer these tiers on an à la carte basis. In
both scenarios, we assume the bundle is no longer offered.22 The allocation of networks into
the theme tiers is described in Table 2.
Assumptions. Before beginning, it is worth mentioning six important assumptions implicit in this analysis. First, as described above, we are not considering the impact to the
potential network’s advertising revenues of being offered à la carte instead of in a bundle.
Many industry participants have argued viewership and advertising revenue will be (possibly substantially) lower under à la carte (Booz Allen Hamilton, 2004; Mandel, 2004).23
Second, we assume that marginal costs to cable systems are positive. This reflects institutional reality: cable networks charge ‘‘affiliate fees’’ (in units of dollars per subscriber per
month) to cable systems for the right to carry that network on a given cable system.24
Third, in the simulations to follow, we must make assumptions about the marginal costs
systems would pay in an à la carte world. For our ‘‘Full À La Carte’’ scenario, we assume
the cable system and programming network share equally in the revenues from à la carte.25
For our ‘‘Theme Tiers’’ scenario, we assume marginal costs to systems are the same as they
are under the bundle. As these marginal costs are important determinants of outcomes and
welfare under à la carte, we discuss the robustness of our results to alternative assumptions
on costs after presenting the baseline results. Fourth, we assume that there are no fixed
costs to the cable system associated with offering an additional channel on either an à
la carte or bundled basis.26 Fifth, we assume that the cable system maximizes short-run
economic profits.27
22
Economic theory has strong predictions about such ‘‘mixed bundling’’: it further increases firm profitability
and reduces consumer welfare (McAfee et al., 1989).
23
The two primary sources of viewer loss often articulated under à la carte are the unwillingness of some
households to purchase a network that they view infrequently and the inability of households to sample new
programming, a typical pre-cursor to frequent viewing.
24
One may argue that the appropriate marginal cost for welfare purposes is zero as that better reflects the
economic cost to society of distributing television networks by satellite. Even so, bundling can still be inefficient:
there can be values of fixed costs exceeding the gross total surplus created by a potential new product for which it
is still profitable for the monopolist to offer it. As for our earlier results, this is more likely the more negatively
correlated are tastes for networks.
25
While contracts between networks and cable systems are sensitive competitive information, industry sources
confirm this is a typical revenue sharing arrangement in these contracts (when included).
26
On institutional grounds, substantial fixed costs to systems are unlikely. For the average cable system, adding
a network requires negotiating a carriage agreement with the network and including it in the programming lineup
of the system. The only potential cost of any consequence would arise if a cable system didn’t have the technical
capability of offering à la carte networks in a low-cost way. Since most cable systems now control access to à la
carte (e.g. Premium) networks using addressable converters, the ‘‘set-top boxes’’ that decompress and unscramble
television signals before passing them to the television, adding a network merely requires re-programming a
system’s converters and is effectively costless. There could be costs to households, however, if they previously did
not use an addressable converter. Furthermore, without addressable converters, there could be substantial
incremental fixed costs of offering à la carte networks.
27
Television content is now becoming available over the Internet (both legally and illegally), introducing the
possibility of disintermediation of cable systems over time. Furthermore, early indications are that this content is
likely to be available on an à la carte (indeed individual program) basis. If cable operators are able to effectively
maintain copyright protection (a big ‘‘if’’), our view is that this represents a form of ‘‘mixed bundling’’ that simply
further increases firm profitability (cf. Footnote 22). That being said, there could be long-run consequences of
bundling by cable systems: it may encourage consumers to seek alternative sources of distribution.
G.S. Crawford, J. Cullen / Information Economics and Policy 19 (2007) 379–404
393
The final assumption relates to competition. The model in this paper is that of a
monopoly cable provider. A typical US cable system, however, competes with two satellite
providers and faces the prospect of further competition from local telephone companies,
internet delivery of television programming, and other distribution technologies.
There are three potential impacts of competition on our analysis. The first relates to the
potential impact of satellite competition on recovering preferences for bundles. We will
momentarily be recovering an estimate of the distribution of household WTP for
Expanded Basic service in Monterey based on an assumption that the system there is a
monopolist. But is it? In a way, it is: satellite systems select their programming and prices
on a national basis, not in response to the products and prices offered (just) in Monterey.
In any local cable market, national satellite pricing implies the local cable system is best
modelled as a monopolist on the residual demand defined by the aggregate demand for
multi-channel video programming less those households that purchase satellite service.28
The assumption in this paper is that households in Monterey purchase either satellite or
cable based on factors uncorrelated with the WTP for the programming offered by California Cable of Monterey.29 We then measure the distribution for WTP for networks and
bundles of those networks for the cable households.
A second issue relates to the impact of competition when networks are offered à la
carte. If all distributors carry all networks, there is little to differentiate a network offered
on cable versus satellite. Competition is therefore likely to be stronger with à la carte
sales.30 What impact would this have on our welfare comparisons of à la carte and bundled sales? If stronger competition reduced prices for individual networks under à la carte,
the result would be increased consumer welfare for those networks that continued to be viable. The analysis in the previous section suggests, however, that fewer networks would be
viable, making the aggregate welfare effects unclear.31
While not the intent, the consequence of these assumptions is to provide a fairly favorable scenario for à la carte pricing. From a policy perspective, this scenario is useful for
two reasons. First, it identifies whether à la carte could possibly improve welfare. Second,
it identifies the areas of uncertainty that must be resolved before à la carte could be considered a viable policy option. While addressing the advertising market and competition in
distribution is beyond the scope of this paper, we can address the consequence of our
assumptions on costs to our conclusions. We do so after presenting the baseline results.
28
Chu (2006) makes a similar assumption.
We will go wrong if satellite consumers are disproportionately selected from some portion of the distribution
of WTP for multi-channel video programming. In practice, this could go either way – at least one of the two US
satellite providers (DirecTV) offers a broader range of programming, including exclusive sports programming,
while the other (Dish Network) offers programming comparable (if not better) than most local cable systems at a
lower price.
30
Bundling likely softens competition by differentiating providers, particularly after accounting for nonprogramming services provided by some distributors (e.g. high-speed internet and telephone service by some cable
systems). Even if an à la carte regime required the purchase of a minimal bundle from a distributor and there are
moderate costs to switching distributors, a network offered on an à la carte basis is likely to be less differentiated
than if it were bundled.
31
The product choice literature emphasizes that competitive markets may provide too many products and there
is some evidence of this in radio markets (Berry and Waldfogel, 1999). This is unlikely to be true for cable
networks, however, as competing distributors share in covering the fixed costs of programming networks rather
than duplicating them.
29
394
G.S. Crawford, J. Cullen / Information Economics and Policy 19 (2007) 379–404
Table 1
Market outcomes and welfare: Baseline Bundle and two À La Carte scenarios
Baseline:
Bundle
Scenario 1:
Full À La Carte
Scenario 2:
Theme Tiers
Market outcomes
Total price
Average cable network price
Average market share (%)
$43.06
$0.86
75.6
$78.54
$1.47
34.1
$50.22
$0.90
44.1
Welfare (all networks viable) per-household
Consumer
Producer
Total
$10.37
$20.52
$30.89
$17.17
$11.44
$28.61
$11.84
$17.29
$29.13
–
$12.31
$6.48
Welfare (considering network exit)
Difference in profit adding network to the bundle
versus scenario
Network exit required to equate consumer surplus
between bundle and scenario
–
41
24
3.1. The baseline results
Our policy simulations are done in two stages. In the first stage, we recover the distribution of household WTP for bundles and marginal costs consistent with observed market
outcomes (prices and market shares) in Monterey. Given that distribution, we then calculate baseline welfare measures (consumer surplus, (gross) producer surplus, and (gross)
total surplus) implied by these preferences and costs. In the second stage, we calculate
market outcomes and welfare measures under ‘‘Full À La Carte’’ and ‘‘Theme Tiers’’,
two policy proposals that have been discussed in the popular press. The functional form
and distributional assumptions underlying each stage are introduced in turn.
3.1.1. The first stage: WTP for bundles and baseline welfare
The most popular Expanded Basic is available to consumers in Monterey at a price of
$44.05 with 72% of households subscribing. What then must be the distribution of WTP
for this service? What are California Cable of Monterey’s marginal costs? To solve for
these objects, we make two assumptions. First, we assume the WTP for bundles of networks are distributed normally with mean lB and variance r2B . In practice, this is a weak
assumption: the distribution of WTP for individual networks could be almost anything
and the resulting WTP for bundles of those networks would be normal.32 Second, we
assume that marginal costs in Monterey are 28% of the price.33 Under these assumptions
(and after taking out $5 for the Limited Basic bundle), we find WTP for Expanded Basic
service in Monterey is normally distributed with mean $47.00 and variance $151.29.
Given this result, we can then calculate the welfare effects of the existing bundle. These
are given in the first column of Table 1. As can be seen there, although we come close to
matching the observed price and market share for Expanded Basic cable service in Mon-
32
The reason: Central Limit Theorems (CLTs). In statistics, CLTs show that the distribution of a sum of
random variables tends to be normally distributed as long as the individual components are not too positively
correlated. See Crawford (2006) for a deeper discussion of this issue applied to bundles of cable networks.
33
This is consistent with industry sources (Halfon, 2003, footnote 78; FCC, 2003).
G.S. Crawford, J. Cullen / Information Economics and Policy 19 (2007) 379–404
395
Fig. 7. Weekly cume for cable networks.
terey, we don’t match them exactly.34 For our predicted outcomes, the average cable subscriber receives a surplus of $13.72 and California Cable earns average gross profits of
$27.14 for a gross total surplus of $40.86. Similar values for the average Monterey household, cable subscriber or not, are $10.37, $20.52, and $30.89.35
3.1.2. The second stage: preferences for networks and À La Carte welfare
To simulate market outcomes and welfare in an à la carte world, we must again make
assumptions about the distribution of WTP and marginal costs, this time for each network
34
For internal consistency, all the counterfactual policy comparisons presented in the paper use our predicted
values as the baseline.
35
Per-household measures prove useful later for comparisons with the à la carte scenarios.
396
G.S. Crawford, J. Cullen / Information Economics and Policy 19 (2007) 379–404
offered in Monterey. We acknowledge up front that there are many possible assumptions
one could make that would yield identical values for lB and r2B that we recovered in the
first stage. We’ve tried to base our assumptions on institutional patterns known in the
industry; estimating these directly from data is the topic of related work in progress (Crawford and Yurukoglu, 2007).
We assume preferences for each network have a very particular form. For each network, j, in Monterey, we assume that only a portion of households, cj, have positive
WTP for that network; the remaining proportion, 1 cj, value it at zero.36 We further
assume that among those that have positive WTP, tastes for each network j are distributed
independently of other networks as a log-normal with location parameter lj and (common) shape parameter, r2.37 We consider the impact of correlation in tastes for networks
after presenting the baseline results.
What of costs? Kagan World Media (2007) reports average affiliate fees for most of the
networks offered by California Cable in Monterey.38 Based on the assumption that marginal costs are 28% of the Expanded Basic price in Monterey (net the $5 for Limited
Basic), this implies California Cable gets a 8.9% discount relative to these national averages.39 Applied to each network this implies its marginal cost = cj = 91.1%*Kaganj. Table
2 reports the average Kagan costs.
What of marginal costs in an à la carte world? There are both institutional and theoretical reasons to believe television networks would negotiate higher prices from cable systems if they are offered on an à la carte basis.40 Furthermore, marketing, operating, and
equipment costs could rise under à la carte sales.41 As described earlier, we assume that
systems share revenue in the ‘‘Full À La Carte’’ scenario and that marginal costs do
not change in the ‘‘Theme Tier’’ scenario.
Under these assumptions, there are two sets of free parameters, the network-specific
means of WTP, lj, and the scale parameter, r2. To pin down lj, we assume mean tastes
for each network is a constant multiple of the (estimated) affiliate fee charged to California
Cable in Monterey, i.e. lj = kcj. This provides enough structure to recover the distribution
of tastes for each network: simply solve for the k and r2 such that the sum of the distributions of WTP for the individual networks equals a normal distribution with mean lB
^2 ¼ 0:23.
and variance r2B . Doing so yielded ^
k ¼ 4:3 and r
36
In our simulations, we assume that cj equals 2.5 times the weekly cumulative viewing of each network from a
given week in 2004. The weekly cumes for cable networks for the first quarter of 2004 are taken from Mandel
(2004) and shown in Fig. 7.
2
2
2
37
The mean and variance of a log-normal with these parameters is EX j ¼ elj þr =2 and VX j ¼ ðer 1Þe2lþr .
38
Networks offered that did not have affiliate fee data were Headline News, Telemundo, California Channel,
KMWB, and the shopping networks (Home Shopping Network, QVC, and ShopNBC). We assumed a low WTP
for shopping networks given that the networks generate most of their revenue from advertising and probably are
provided to systems at low or zero cost. For the remaining, we assigned to them the median cost of a cable
network. If anything, this likely overstates their actual costs.
39
This is not surprising. Comcast (the owner of California Cable) is the largest operate of cable systems in the
US and is known to be an aggressive negotiator with program networks.
40
Kagan Media Research (2005) ‘‘estimates TV channel operators would need to raise per-capita channel
carriage fees by a multiple of four to offset a 50% loss of subscribers from big basic bundles’’. This general pattern
of higher affiliate fees for à la carte networks is an equilibrium outcome of a model by Rennhoff and Serfes (2005).
41
For example, marketing costs for (à la carte) Premium cable networks are in the range of 15–25% of sales,
while those for (bundled) Expanded Basic cable networks are 2–6% of sales (Booz Allen Hamilton, 2004, Figure
16). Similarly, bundling (or alternatives like subscriber-selected tiers) are much less complex, saving on costs (Chu
et al., 2006; Wildman, 2006).
G.S. Crawford, J. Cullen / Information Economics and Policy 19 (2007) 379–404
397
Table 2
Theme tiers and networks costs
Tier
Network name
National average network cost
Sports
ESPN
ESPN 2
Golf Channel
Outdoor Life Network
Fox Sports Net
$2.28
$0.21
$0.20
$0.11
$1.34
Family
Nickelodeon
ABC Family Channel
Cartoon Network
Disney Channel
Hallmark Channel
$0.38
$0.21
$0.14
$0.76
$0.03
General Interest
Turner Network TV
USA Cable
Lifetime
Spike TV
TBS Superstation
FX
TV Land
oh! Oxygen
$0.82
$0.44
$0.20
$0.17
$0.34
$0.32
$0.08
$0.05
News and Public Affairs
CNN
Weather Channel
CNBC
Headline News
Fox News Channel
MSNBC
C-SPAN
C-SPAN 2
$0.43
$0.09
$0.25
$0.09*
$0.22
$0.14
$0.05
$0.05
Movie
Arts & Entertainment
American Movie Classics
$0.20
$0.22
Music and Entertainment
MTV
VH1
BET
Bravo
Country Music TV
$0.26
$0.12
$0.13
$0.13
$0.04
Spanish
GalaVision
Telemundo
$0.03
$0.09*
Shopping
Home Shopping Network 2
QVC
ShopNBC
$0.01*
$0.01*
$0.01*
Science and Nature
Discovery Channel
Learning Channel
Animal Planet
$0.24
$0.15
$0.07
Specialty
History Channel
Comedy Central
HGTV
E! Entertainment Television
Food Network
$0.17
$0.10
$0.06
$0.19
$0.06
(continued on next page)
398
G.S. Crawford, J. Cullen / Information Economics and Policy 19 (2007) 379–404
Table 2 (continued)
Tier
Network name
National average network cost
Court TV
$0.10
Other
TV Guide Channel
California Channel
KMWB (WBN) Minneapolis-St. Paul
$0.03
$0.09*
$0.09*
Notes: Reported are the allocation of cable networks into theme tiers for use in the ‘‘Theme Tier’’ scenario
(cf. Table 1) and the national average per-subscriber cost for that network from Kagan World Media (2007).
Costs for starred networks are estimated; see Footnote 38 for details.
3.1.3. Scenario 1: Full À La Carte
We first consider our results from full à la carte pricing. Prices for channels ranged from
a high of $9.34 for ESPN to a low of $0.22 for GalaVision (a Spanish-language network),
with an average price per network of $1.47 and an average market share of 34.1%.42 Table
3 lists market outcomes and estimated per-household welfare for each of the channels
offered in Monterey. If a consumers were to buy all the channels previously offered in
the bundle, the total price would be $78.54 (including the $5 for ‘‘Limited Basic’’), slightly
less than double the bundled price.
What are the welfare consequences of full à la carte? We answer this question in two
parts. We first answer assuming that each of the networks offered in Monterey continues
to be offered. Under this assumption, the second column of Table 1 shows that consumers
are (much) better off and firms are (much) worse off, at a moderate cost to society. Average
per-household consumers surplus under à la carte is $17.17 (a 65.6% increase), gross firm
profit is $11.44 (a 44.2% decrease), and gross total surplus is $28.61 (a 7.4% decrease). If
our assumptions are accurate, it is no wonder firms are adamantly opposed to à la carte.
In Section 2.2, we demonstrated, however, that because bundling is more profitable
than à la carte sales, some channels may no longer be viable under à la carte. Similarly,
there may be fixed costs to cable systems that exceed their profit from a network offered
on an à la carte basis. We address this issue in two ways. First, we calculate the profit of
each network from à la carte sales versus the incremental profit that adding the network
would bring to a bundle containing all of the other networks offered in Monterey. As
expected, incremental per-household bundling profits are on average $12.31 higher
(107.4% of per-household à la carte profits), suggesting that there could be values of fixed
costs such that networks would be offered under bundling but not under à la carte sales.
Could these losses outweigh the aggregate consumer surplus gains from à la carte? To
address this question, we rank networks by their à la carte profit and calculate how many
would have to exit from the market in order for à la carte to yield surplus to the average
consumer no higher than they get from purchasing the bundle. The results are quite stark.
Fully 41 of the 50 offered cable networks would have to exit for consumers surplus under
full à la carte to be no higher than in a bundled world.43 While some networks would
surely exit, it seems unlikely that so many would, suggesting consumer welfare would
likely be higher under ‘‘Full À La Carte’’ than bundled sales.
42
These calculations and those for Theme Tiers subtract off the $5.00 for Limited Basic when calculating the
average price per network or price per tier.
43
The last four networks that would have to be dropped are estimated to be AMC, The Discovery Channel,
CNBC, and MTV.
G.S. Crawford, J. Cullen / Information Economics and Policy 19 (2007) 379–404
399
Table 3
Full À La Carte simulation results
Price
Discovery Channel
ESPN
Turner Network TV
USA Cable
Arts & Entertainment
CNN
Lifetime
Nickelodeon
Spike TV
TBS Superstation
Weather Channel
ABC Family Channel
CNBC
ESPN 2
Headline News
History Channel
Learning Channel
MTV
VH1
American Movie Classics
Cartoon Network
Comedy Central
Fox News Channel
HGTV
Animal Planet
E! Entertainment Television
Disney Channel
Food Network
FX
TV Land
MSNBC
Court TV
BET
Bravo
Country Music TV
TV Guide Channel
Golf Channel
Hallmark Channel
Outdoor Life Network
Oh! Oxygen
C-SPAN
C-SPAN 2
California Channel
Fox Sports Net
GalaVision
Home Shopping Network 2
KMWB
QVC
ShopNBC
Telemundo
$0.87
$9.34
$2.27
$1.27
$0.78
$2.12
$0.77
$1.23
$0.71
$0.93
$0.42
$0.93
$3.07
$1.09
$0.70
$0.74
$0.63
$1.04
$0.53
$0.93
$0.64
$0.45
$1.24
$0.34
$0.38
$0.79
$3.06
$0.36
$1.24
$0.47
$0.99
$0.69
$1.10
$0.83
$0.37
$0.23
$7.72
$0.30
$2.21
$0.54
$5.18
$5.44
$0.64
$5.44
$0.22
$0.35
$0.64
$0.35
$0.36
$0.64
Market share (%)
53
47
69
66
49
39
50
59
46
69
41
43
15
37
24
44
46
48
43
45
42
42
34
34
35
46
47
32
49
32
27
28
23
30
21
25
5
19
9
18
2
2
27
47
27
5
27
5
5
27
Per household surplus
Cost
Consumer
Producer
Total
$0.34
$3.24
$1.17
$0.62
$0.29
$0.62
$0.29
$0.55
$0.24
$0.48
$0.13
$0.30
$0.36
$0.30
$0.13
$0.24
$0.21
$0.37
$0.17
$0.32
$0.20
$0.14
$0.31
$0.09
$0.10
$0.27
$1.09
$0.09
$0.46
$0.11
$0.20
$0.14
$0.19
$0.19
$0.06
$0.04
$0.28
$0.04
$0.15
$0.07
$0.07
$0.07
$0.13
$1.95
$0.04
$0.01
$0.13
$0.01
$0.01
$0.13
$0.23
$2.17
$0.78
$0.42
$0.19
$0.41
$0.19
$0.36
$0.16
$0.32
$0.09
$0.20
$0.24
$0.20
$0.09
$0.16
$0.14
$0.25
$0.11
$0.21
$0.13
$0.10
$0.21
$0.06
$0.07
$0.18
$0.73
$0.06
$0.30
$0.08
$0.13
$0.10
$0.12
$0.12
$0.04
$0.03
$0.19
$0.03
$0.10
$0.05
$0.05
$0.05
$0.09
$1.28
$0.03
$0.01
$0.09
$0.01
$0.01
$0.09
$0.57
$5.41
$1.95
$1.04
$0.48
$1.03
$0.48
$0.91
$0.41
$0.81
$0.22
$0.50
$0.60
$0.50
$0.21
$0.41
$0.36
$0.62
$0.28
$0.53
$0.33
$0.23
$0.52
$0.14
$0.17
$0.45
$1.81
$0.14
$0.76
$0.19
$0.33
$0.24
$0.31
$0.31
$0.10
$0.07
$0.47
$0.07
$0.26
$0.12
$0.12
$0.12
$0.22
$3.23
$0.07
$0.02
$0.21
$0.02
$0.02
$0.22
$0.43
$4.67
$1.13
$0.64
$0.39
$1.06
$0.38
$0.61
$0.35
$0.47
$0.21
$0.47
$1.54
$0.54
$0.35
$0.37
$0.31
$0.52
$0.26
$0.47
$0.32
$0.23
$0.62
$0.17
$0.19
$0.39
$1.53
$0.18
$0.62
$0.23
$0.50
$0.34
$0.55
$0.41
$0.19
$0.11
$3.86
$0.15
$1.10
$0.27
$2.59
$2.72
$0.32
$2.72
$0.11
$0.18
$0.32
$0.18
$0.18
$0.32
400
G.S. Crawford, J. Cullen / Information Economics and Policy 19 (2007) 379–404
Table 4
Theme Tier simulation results
Price
Family
General Interest
Movie
Music & Entertainment
News & Public Affairs
Other
Science & Nature
Shopping
Spanish
Specialty
Sports
$5.65
$7.62
$1.72
$2.55
$5.10
$1.03
$1.71
$0.38
$0.77
$2.42
$16.28
Market share (%)
51
75
44
48
43
34
50
14
26
54
46
Per household surplus
Cost
Consumer
Producer
Total
$1.43
$2.19
$0.42
$0.68
$1.47
$0.23
$0.45
$0.04
$0.13
$0.64
$4.15
$2.16
$4.07
$0.58
$0.93
$1.66
$0.29
$0.65
$0.05
$0.17
$0.98
$5.76
$3.59
$6.26
$1.00
$1.61
$3.13
$0.52
$1.10
$0.09
$0.30
$1.62
$9.91
$1.38
$2.21
$0.38
$0.62
$1.20
$0.19
$0.42
$0.03
$0.11
$0.62
$3.77
3.1.4. Scenario 2: Theme Tiers
What of theme tiers? Our results show that these provide intermediate benefits. Assuming marginal costs are the same here as under the full bundle, prices ranged from a high of
$16.28 for the Sports Tier (containing ESPN, ESPN2, Fox Sports Net, the Golf Channel,
and the Outdoor Life Network) to a low of $0.38 for the Shopping Tier (containing QVC,
ShopNBC, and the Home Shopping Network 2). The average price per tier (network) is
$4.11 ($0.90) and the average (tier) market share is 44.1%. Table 4 lists market outcomes
for each of the theme tiers. If a consumers were to buy all the channels previously offered
in the bundle, the total price would be $50.22 (including the $5 for ‘‘Limited Basic’’), only
slightly more than the bundled price.
What are the welfare consequences in this case? Assuming each of the networks offered
in Monterey continues to be offered, the third column of Table 1 shows that consumers are
now moderately better off and firms are moderately worse off. Average per-household consumers surplus under theme tiers is $11.84 (a 14.2% increase), gross firm profit is $17.29 (a
15.7% decrease), and gross total surplus is $29.13 (a 5.7% decrease). As before, consumers
benefit by less than firms lose, yielding a decrease in total surplus.
As for full à la carte, adding networks to the full bundle produces more profit than adding them to theme tiers (by $6.48 or 37.5% of theme tier profits). Again ranking networks
by their à la carte profit now shows 24 of the 50 offered cable networks would have to exit
for consumers surplus under theme tiers to be no higher than in a bundled world.44 Here,
too, it seems unlikely so many networks would no longer be viable, suggesting consumer
welfare would still be higher under theme tiers than the full bundle.
3.2. Robustness
In this subsection we consider the robustness of our results to several of our
assumptions.
44
The last four networks that would have to be dropped to equate welfare in this case are estimated to be Court
TV, Outdoor Life Network, VH1, and Bravo.
G.S. Crawford, J. Cullen / Information Economics and Policy 19 (2007) 379–404
401
3.2.1. Marginal costs
Our assumptions on marginal costs are both qualitatively and quantitatively important
to our results. If, for example, marginal costs to cable systems were to increase by a factor
of three when all networks are offered à la carte, both consumers surplus and the monopolist’s profit would fall, decreasing total surplus by almost a third!
How likely is this scenario? Probably not very. Among all the assumptions on costs
we’ve tried, 50% revenue sharing yielded the greatest aggregate gross profit to cable networks and cable systems next to the bundle. While detailed conclusions therefore depend
on the likely form of contracts between network providers and cable systems in any à la
carte scenario, some hybrid of low marginal costs and some bundling (like our Theme
Tiers) or high marginal costs/revenue sharing (like our Full À La Carte) seems likely.
3.2.2. Correlation
In Section 2, we emphasized how important correlation can be for the profit and welfare consequences of bundling. In our baseline results, however, we assume independence
in tastes for networks. In part, this is of necessity: there are over 1200 correlations between
pairs of 50 networks and there are no empirical results to guide the choice of their values
(or even their signs in some cases). As a very crude robustness test, we re-estimated our
model assigning what we felt were plausible values for correlations both within and
between the networks according to their allocation to our theme tiers and the results were
nearly identical to the baseline specification presented above.
Does that mean correlation is unimportant? No. As noted in Section 2, bundling profitability is increasing in the negative correlation between a new product and an existing
bundle, giving a monopolist strong incentives to carry such networks. This increased profitability also gives the monopolist incentives not to carry networks that are similar to other
networks already included in the bundle (i.e. those for which correlation is likely to be
positive). This factor has, if anything, the opposite effect for à la carte sales; carrying a
competing network could enhance a cable system’s bargaining power with incumbent networks. Bundling is therefore likely to bias cable systems against carrying networks that
compete with established incumbent networks. Indeed, this conjecture can possibly explain
consumers increased frustration about having to pay large annual price increases for cable
bundles when the incremental additions to the bundle are highly-specialized niche networks (Squeo and Flint, 2004). If such networks help attract new subscribers but provide
little value to existing subscribers, they are likely to be negatively correlated with tastes for
the existing bundle. If marginal costs to cable systems for these networks are also above
mean tastes – a possibility if only a small portion of the population values the new networks – the addition of these networks may be profitable but welfare-reducing.45 Evaluating correlation in household tastes for networks are therefore an important area of
future research.
4. Conclusions
In this paper, we conduct a numerical analysis of bundling’s impact on a monopolist’s
pricing and product choice and assess the implications for consumer welfare in cable
45
Even without correlation, we estimate the incremental consumers surplus from bundling to be positive for
only 7 of Monterey’s 50 networks and the incremental total surplus to be positive for only 28.
402
G.S. Crawford, J. Cullen / Information Economics and Policy 19 (2007) 379–404
television markets. Since theoretical models are ambiguous about the net impact of bundling on welfare, we take a numerical approach. We calibrate a model of the pricing decisions of an average monopoly cable system and assess the implications of the model for
consumer welfare in cable television markets.
We have three main findings. First, because bundling aggregates household tastes for
bundle components (here cable networks), it can increase firm profits and reduce consumer
surplus when marginal costs are low relative to mean tastes and when preferences are not
too heterogeneous. Both of these outcomes are likely to be true in cable markets. Second,
across a wide range of specifications, we find that bundling increases the profits of cable
systems relative to à la carte sales, providing stronger incentives for cable systems to offer
networks. Finally, and most important, our simulation results suggest the first effect is
stronger than the second. In particular, in a ‘‘Full À La Carte’’ world, if all networks continue to be offered, average per-household consumers surplus is estimated to increase by
$6.80 (a 65.6% increase), gross firm profit is estimated to fall by $9.08 (a 44.2% decrease),
and gross total surplus is estimated to fall by $2.28 (a 7.4% decrease). These effects almost
surely outweigh any compensating benefit of à la carte to enhance incentives to offer networks; we estimate that 41 of the 50 offered cable networks in Monterey would have to exit
for consumers surplus under Full À La Carte to be no higher than that under bundling.
Theme tiers offer similar, though smaller, effects (a 14.2% increase in consumers surplus,
a 15.7% decrease in profit, a 5.7% decrease in total surplus, and 24 networks needed to exit
to equate consumer benefits). On balance, we find that consumers might well benefit from
à la carte sales of cable networks.
These conclusions come with important caveats. They depend critically on the nature of
fixed and marginal costs to cable systems in an à la carte world, conditions we speculate on
but cannot know. Furthermore, we don’t directly model the consequences of competition
between cable and satellite systems, nor the impact to the advertising market of à la carte
offerings. The latter in particular could tip the scales in favor of bundling. Even more
important, however, is that we have shown that many things can happen under bundling.
To know what will happen requires better information, particularly about consumer preferences for individual television networks, cable systems’ costs under both à la carte and
bundled sales, and the link between household and advertising markets. These are all fruitful topics for further research.
Acknowledgements
We would like to thank the editor, two anonymous referees, Leslie Marx, Scott Wallsten, and Ken Wilbur for helpful comments.
References
Adams, W., Yellen, J., 1976. Commodity bundling and the burden of monopoly. Quarterly Journal of Economics
90, 475–498.
Anderson, S., Coate, S., 2005. Market provision of broadcasting: a welfare analysis. Review of Economic Studies
72, 947–972.
Armstrong, M., 1996. Multiproduct non-linear pricing. Econometrica 64 (1), 51–75.
Armstrong, M., 1999. Price discrimination by a multi-product firm. Review of Economic Studies 66, 151–168.
Bakos, Y., Brynjolfsson, E., 1999. Bundling information goods: pricing, profits, and efficiency. Management
Science 45 (2), 1613–1630.
G.S. Crawford, J. Cullen / Information Economics and Policy 19 (2007) 379–404
403
Berry, S., Waldfogel, J., 1999. Free entry and social inefficiency in radio broadcasting. Rand Journal of
Economics.
Booz Allen Hamilton (2004). The À-La-Carte Paradox: Higher Consumer Costs and Reduced Programming
Diversity, Discussion paper, Booz Allen Hamilton, Report prepared for the National Cable & Telocommunications Association.
Chu, C.S., 2006. ‘‘The Effects of Satellite Entry on Product Quality for Cable Television’’, mimeo. Stanford
University.
Chu, C.S., Leslie, P., Sorensen, A., 2006. ‘‘Incomplex Alternatives to Mixed Bundling’’, mimeo. Stanford
University.
Crawford, G., 2006. ‘‘The Discriminatory Incentives to Bundle in the Cable Television Industry’’, mimeo.
University of Arizona.
Crawford, G., Yurukoglu, A., 2007. ‘‘Demand, Pricing, and Bundling in Subscription Television Markets’’,
mimeo. University of Arizona.
Doyle, C., 1998. Programming in a competitive broadcasting market: entry, welfare and regulation. Information
Economics and Policy 10, 23–39.
Evans, D.S., Salinger, M., 2005. Why do firms bundle and tie? Evidence from competitive markets and
implications for Tying Law. Yale Journal on Regulation 22 (1), 37–89.
Fang, H., Norman, P., 2005. ‘‘To Bundle or Not To Bundle’’, mimeo. Yale University.
FCC, 2003. 2002 Report on Cable Industry Prices’’, Discussion paper, Federal Communications Commission.
Available at http://www.fcc.gov/mb/csrptpg.html.
FCC, 2004. Report on the Packaging and Sale of Video Programming to the Public, Discussion paper, Federal
Communications Commission, November 18, 2004. Available at http://www.fcc.gov/mb/csrptpg.html.
FCC, 2005a. 2004 Report on Cable Industry Prices, Discussion paper, Federal Communications Commission.
Available at http://www.fcc.gov/mb/csrptpg.html.
FCC, 2005b. Eleventh Annual Report on the Status of Competition in the Market for the Delivery of Video
Programming (2004 Report), Discussion paper, Federal Communications Commission, Available at http://
www.fcc.gov/mb/csrptpg.html.
FCC, 2006. Further Report on the Packaging and Sale of Video Programming to the Public, Discussion paper,
Federal Communications Commission, February, 2006. Available at http://www.fcc.gov/mb/csrptpg.html.
Halfon, J., 2003. The failure of cable deregulation: a blueprint for creating a competitive, pro-consumer cable
television marketplace, Discussion paper, United States Public Interest Research Group.
Kagan Media Research, 2005. À La Carte Pricing Makes Great Theory, But TV Channel Bundling Tough to
Beat, Discussion paper, Kagan Media Research. Available at http://www.ncta.com/IssueBrief.aspx?contentId=15.
Kagan World Media, 2007. Economics of Basic Cable Television Networks, Discussion paper, Kagan World
Media.
Mandel, J., 2004. Cable À La Carte or Basic: An Advertiser Perspective, Discussion paper. MediaCom,
Comments available at http://www.fcc.gov/realaudio/tr072904.pdf.
McAfee, R.P., McMillan, J., Whinston, M., 1989. Multiproduct monopoly, commodity bundling, and correlation
of values. Quarterly Journal of Economics 103, 371–383.
Nalebuff, B., 2004. Bundling as an entry barrier. Quarterly Journal of Economics 119 (1), 159–187.
Owen, B., Wildman, S., 1992. Video Economics. Harvard University Press.
Rennhoff, A., Serfes, K., 2005. ‘‘The Role of Upstream-Downstream Competition on Bundling Decisions: Should
Regulators Force Firms to Unbundle?’’, mimeo. Drexel University.
Reuters, 2003. U.S. Lawmaker Urges A La Carte Cable Channel Rates. Reuters News Service.
Rochet, J.-C., Chone, P., 1998. Ironing, sweeping, and multidimensional screening. Econometrica 66 (4), 783–
826.
Rochet, J.-C., Stole, L., 2000. ‘‘The Economics of Multidimensional Screening’’, mimeo. University of Chicago.
Salinger, M., 1995. A graphical analysis of bundling. Journal of Business 68 (1), 85–98.
Schmalensee, R., 1984. Gaussian demand and commodity bundling. Journal of Business 62, S211–S230.
Shatz, A., 2006. À la Carte Pricing May Cut Bills For Cable Customers, FCC Says. Wall Street Journal.
Spence, A.M., 1975. Monopoly, quality, and regulation. Bell Journal of Economics 6, 417–429.
Squeo, A.M., Flint, J., 2004. Should Cable Be A la Carte, Not Flat Rate? Wall Street Journal.
Steiner, P., 1954. Program patterns and preferences and the workability of competition in radio broadcasting.
Quarterly Journal of Economics 66, 194–223.
404
G.S. Crawford, J. Cullen / Information Economics and Policy 19 (2007) 379–404
Stigler, G., 1963. United States v. Loew’s Inc.: A Note on Block Booking. In: Kurland, P. (Ed.), The Supreme
Court Review. University of Chicago Press, pp. 152–157.
Stigler, G., 1968. A Note on Block Booking. In: Stigler, G., Irwin, Richard D., (Eds.), The Organization of
Industry.
Tirole, J., 1988. The Theory of Industrial Organization. MIT Press.
Warren (Ed.), 2005. Television and Cable Factbook. Warren Publishing, Inc.
White, H., 1984. Asymptotic Theory for Econometricians. Academic Press.
Wildman, S., 2006. A Case for A La Carte and Increased Choice? An Economic Assessment of the FCCs Further
Report, Discussion paper, Michigan State University.